More QTL for owering time revealed by substitution lines in Brassica oleracea

Transcription

1 Heredity 83 (1999) 586±596 Received 17 March 1999, accepted 27 May 1999 More QTL for owering time revealed by lines in Brassica oleracea A. M. RAE, E. C. HOWELL & M. J. KEARSEY* Plant Genetics and Cell Biology Group, School of Biological Sciences, The University of Birmingham, Birmingham B15 2TT, U.K. Seventy-nine recombinant backcross lines from a cross between Brassica oleracea var. italica and Brassica oleracea var. alboglabra were grown in eld trials over ve years along with the alboglabra recurrent parent. Plants were scored for the days from sowing to the opening of the rst ower, and lines that owered signi cantly earlier or later than the recurrent parent were identi ed. Based on the lengths of the s, evidence for 11 QTL on chromosomes O1, O2, O3, O5 and O9 was found, ve of which mapped to similar regions to ve of the six found in a previous analysis of doubled haploid lines from the same cross. Several of the QTL were linked closely in repulsion. Keywords: Brassica, owering time, QTL analysis, lines. Introduction Brassica oleracea is an agriculturally important species including many vegetables such as cabbage, broccoli and cauli ower. It is closely related to the model dicotyledonous plant Arabidopsis thaliana and so it is expected that information concerning control of basic biological processes in A. thaliana will be transferable to Brassica crops. Information should also be transferable between di erent species within the Brassica genus, as the three diploid genomes, A, B and C, of B. rapa, B. nigra and B. oleracea, respectively, are thought to have derived from the same ancestor. These genomes reveal striking conservation of content, although chromosome duplication and translocation have occurred during divergence (Lagercrantz & Lydiate, 1996). The amphidiploid B. napus (oilseed rape) genome is made up of the A and C genomes probably from close relatives of B. rapa and B. oleracea. (U, 1935; Parkin et al., 1995). Flowering time is not only of scienti c interest because it facilitates the understanding of plant development, it is also important in agriculture, because its modi cation may enable the geographical range of the Brassica crop to be extended. For example, cultivation of B. napus is usually restricted to temperate latitudes, but the development of early owering cultivars would allow the crop to be grown in low-rainfall regions such as the Western Australian wheat belts (Thurling & *Correspondence. m.j.kearsey@bham.ac.uk Depittayanan, 1992), and more northern regions of Canada (Murphy & Scarth, 1994). Genes a ecting owering time have been identi ed in A. thaliana by mutagenesis (Koornneef et al., 1991). For example, mutations of the CONSTANS gene cause delayed owering under long days, but not short, and the gene has been cloned by chromosome walking (Putterill et al., 1993, 1995). Two regions in uencing owering time in B. nigra (on LG2 and LG8) have been found to be homologous to the CONSTANS gene region (Lagercrantz et al., 1996). These regions have also been shown to carry quantitative trait loci (QTL) in the A genome of B. napus, and show large-scale colinearity between regions on chromosomes O2, O3 and O9 of B. oleracea. Previous studies mapping QTL for owering time in B. oleracea have been reported. For example, Kennard et al. (1994) found signi cant QTL using single-factor ANOVA on an F 2 from a cabbage broccoli cross, whereas Camargo & Osborn (1996) used F 3 families from a di erent cross between cabbage and broccoli. Because of the lack of both a standardized nomenclature for linkage groups in Brassica species and of common probes, it is not possible to compare the QTL locations found in these studies. Another drawback to these studies is the limited life-span of the populations used. The heterozygosity of individuals prevents their maintenance by sel ng, so trials can only be carried out once in a single environment. Bohuon et al. (1998) overcame this by using doubled haploid lines derived from the F 1 of a cross between a Chinese kale and a hybrid of 586 Ó 1999 The Genetical Society of Great Britain.

2 QTL FOR FLOWERING TIME 587 calabrese, resulting in six QTL for owering time being mapped to chromosomes O2, O3, O5 and O9. The use of doubled haploid plants still has the limitation that segregating populations can map QTL only to a relatively large region of a chromosome. This present paper reports work which reinforces the results of Bohuon et al. (1998) by using the same cross to engineer near-isogenic lines. Repeated backcrossing and sel ng of the progeny from this cross have enabled the production of lines that contain small segments of donor parent DNA within a recurrent parent background (Ramsay et al., 1996). Di erences in phenotype between the lines and the recurrent parent should be caused by the substituted region. The lines should be homozygous so that the lines can be maintained and multiplied by sel ng allowing the same lines to be grown in di erent environments so that experiments can be repeated over a number of years and conditions. The precision to which a QTL can be mapped depends upon the size of the introgressed region, but comparisons between di erent lines can narrow these regions further. Materials and methods Production of experimental lines The parent plants were descended from microsporederived doubled haploid lines. The recurrent parent is A12DHd, derived from Brassica oleracea var. alboglabra, and the donor is GDDH33, which is Brassica oleracea var. italica, derived from a commercial calabrese F 1 hybrid variety, Green Duke (Bohuon et al., 1996). For simplicity, the doubled haploid lines used as parents will be referred to as A12 and GD, respectively. The lines were produced (as described in Ramsay et al., 1996) by using a single F 1 (A12 GD) to pollinate three A12 plants to produce the rst backcross generation (BC 1 ). A12 was used as the female to ensure that all cytoplasm from the rst backcross generation and subsequent generations was A12 in origin. Lines were selected for a high number of purely recurrent linkage groups, wide coverage of the genome by the donor introgression and duplications of introgressed regions. Lines were backcrossed again and then selfed two or three times to produce lines with homozygous regions of introgressed GD DNA in an A12 background. Seventy-eight RFLP probes selected from those previously described by Bohuon et al. (1996) to construct a linkage map from doubled haploid lines produced from the F 1 of this cross, were used to map the substituted regions in the lines together with an additional 119 AFLP markers (Sebastian et al. 1999). This gave an average spacing between markers of 4.5 cm. The positions of the introgressed GD DNA in the 79 lines are shown in Appendix 1. The minimum indicates the region known to be GD DNA, whereas the maximum includes the region between A12 and GD markers in which recombination has occurred. It is not yet possible to de ne the proportion of this region which is GD, so the whole section must be taken into account. Trial plan Eight eld trials were carried out over a period of ve years; one each in 1994 and 1995, two each in 1996, 1997 and Lines were used in trials as they became available from the breeding programme. Over the ve years, 79 lines, with varying numbers of replicates, were grown alongside the recurrent and the donor parents. Table 1 shows the sowing dates and the number of lines and replicates for each trial. Seeds were sown in 5 cm plastic pots, using John Innes compost no. 3, in the glasshouse and placed in a single block with complete individual plant randomization, surrounded by guard plants. This design was used to maximize the power to detect QTL (Kearsey & Pooni, 1996). The glasshouse was unheated and unlit so that temperature and daylength depended on natural conditions. After nine days the number of germinated seeds was noted, and plants were thinned from two to one per pot. Any pots in which no seed germinated were replaced with guard plants from the same line. The pots were moved to an open-ended polythene tunnel after 15 days for acclimatization. After 23 days the seedlings were planted in the eld in the same random position. The eld was set out in rows of 100 plants, Table 1 The sowing dates for the eight Brassica oleracea trials over the ve years, with the number of lines and the number of replicates. A12 and GD are the parent lines and SL denotes the lines Year Trial Sowing date A12 GD SL 1998 T1 12/05/98 A GD SL T2 26/05/98 A GD SL T1 13/05/97 A GD SL T2 27/05/97 A GD SL T1 20/05/96 A Ð 8 SL T2 12/06/96 A GD SL T1 09/06/95 A GD SL T1 01/06/94 A GD SL 20

3 588 A. M. RAE ET AL. spaced 76 cm between rows and 31 cm between plants in the rows. Guard plants were grown around the perimeter and in the spaces left by dead plants. The whole trial area was protected by nylon netting to prevent bird damage. Plants were irrigated during dry periods, and chemical treatments were used to control pests. Data collection Flowering was checked daily, and scored as the number of days from sowing to the rst ower opening on each plant. Data from plants that were badly a ected by cabbage root y or grey aphid were removed. Other traits were also recorded, but their analysis will be published elsewhere. Data analysis 1994, 1995 and 1996 trials Four one-way analyses of variance (ANOVAs) were carried out to test for variation between lines within each trial in 1994, 1995 and The within-lines mean squares from these ANOVAs were used to carry out Tukey±Kramer multiple comparison tests (Dunnett, 1980), to determine which lines owered signi cantly earlier or later than the A12 recurrent parent trials Seeds for each line in both trials in 1997 were taken from two di erent parent plants which had been grown in two completely randomized blocks in glasshouses. This was to allow maternal e ects in the progeny to be estimated. The two eld trials contained the same number of plants from the same parent plants so that variation between trials could also be examined. The data for both trials were standardized to mean zero and variance one, to minimize genotype-byenvironment e ects arising from di erent ranges in owering time, so allowing the trials to be analysed together. A cross-classi ed hierarchical ANOVA was carried out for the lines on the standardized data to test for interactions between genotypes and environments. This was repeated for the A12 parents in the two trials. The within-lines mean square for these ANOVAs were combined for use in a Tukey±Kramer multiple comparison test trials The lines used in the 1998 eld trials were selected as being earlier or later owering than A12 based on the analysis of lines in previous years. This allowed the use of one-tailed tests to compare each line against the recurrent A12 parent. The data from each trial were standardized permitting the trials to be analysed together. Results 1994, 1995 and 1996 trials The ANOVAs for the four trials in 1994, 1995 and 1996 indicated signi cant variation between lines within trials, implying that the di erent s of GD a ected the owering time of the lines. The Tukey±Kramer test showed the lines SL112, SL113, SL140, SL175 in 1994; SL113, SL129, SL140, SL172 in 1995; SL102 in 1996 trial 1; and SL102, SL128, SL137, SL157, SL172, SL175 and SL178 in 1996 trial 2 were signi cantly later owering (at the 5% level) than the A12 parent. Line SL108 was signi cantly earlier owering than A12 in data The mean and standard deviation for owering time for trial 1 and trial 2 (x ˆ , s ˆ and x ˆ , s ˆ 4.041, respectively) were used to standardize the data. The cross-classi ed hierarchical ANOVA (Table 2a) showed that the variation between lines was highly signi cant (P < 0.001), and that replicate families within lines also di ered (0.01 > P > 0.05). The latter indicates the importance of maternal e ects. The interaction of lines with trials, and the families within lines interaction with trials, were not signi cantly di erent from each other (1.67/1.12 ˆ 1.49, P > 0.05), although families within lines did interact with trials signi cantly when compared to the error (P < 0.01). As expected, the variation between trials was very small (P > 0.05), any di erence from zero being attributed to unequal numbers of observations in each trial. A similar ANOVA was carried out for A12 recurrent parent plants (Table 2b), showing that the variation between di erent families within the A12 line and the variation between trials were not signi cant. However, interaction between trials and families was just signi cant despite the scalar transformation. In order to carry out the Tukey±Kramer multiple comparison, the mean square `between families within lines' combined with the mean square `between families' for the A12 was used as the error variation. The Tukey±Kramer test showed that lines SL115, SL134, SL141 and SL175 were signi cantly later owering than the A12 recurrent parent data Flowering data for the two trials in 1998 were standardized, using the mean and standard deviation for the two trials (x ˆ , s ˆ and x ˆ , s ˆ 3.551,

4 QTL FOR FLOWERING TIME 589 Table 2 (a). The cross-classi ed hierarchical ANOVA for the Brassica oleracea lines in 1997 trials 1 and 2. The F-value is taken to be nonsigni cant (NS) when P > 0.05; *indicates 0.05 > P > 0.01, **indicates 0.01 > P > 0.001; ***indicates P < (b). The cross-classi ed ANOVA for the A12 recurrent parent in 1997 trials 1 and 2 (a) Item d.f. Item d.f. SS MS F P Trial 1 Trial NS Families 77 Lines *** Families/Lines * Trial Families 77 Lines Trials NS (Families/Lines) Trials ** Error 2209 Error Total 2364 Total (b) Item d.f MS F P Trial NS Family NS Trial Family * Error Total 276 respectively). ANOVAs were carried out using the combined data from the two trials. Lines SL115, SL121, SL127, SL133, SL134, SL138, SL141, SL142, SL158, SL172, SL175 and SL177 were shown to be signi cantly later owering than the A12 parent, whereas SL108 and SL122 were signi cantly earlier owering. Of the 23 lines that showed signi cant variation in owering behaviour from the A12 parent, 10 have a single, four have two s, eight have three regions and one has four regions substituted with GD. The following interpretation should be read in conjunction with Fig. 1 Signi cant lines Table 3 shows the mean of the A12 parent in each year, together with the means for each line that di ered signi cantly in owering time from the A12. Over the eight trials it can be seen that only two lines (SL108 and SL122) were signi cantly earlier owering, whereas 21 were signi cantly later owering than the A12 parent (SL102, SL112, SL113, SL115, SL121, SL127, SL128, SL129, SL133, SL134, SL137, SL138, SL140, SL141, SL142, SL157, SL158, SL172, SL175, SL177, SL178). Discussion Seventy-nine lines were grown over the ve years. Of these, only two lines showed evidence for earlier owering than A12, whereas 21 lines were signi cantly later owering. This bias was to be expected because sections of chromosomes from the late owering GD parent had been substituted into the early owering A12 recurrent parent. Lines with one Any variation between the A12 and the 10 lines containing just one must result from GD DNA in this region, therefore such lines enable the GD DNA region to be treated as a single segregating factor. Because it is not possible to de ne the proportion of GD DNA in the region in which recombination has occurred, it was assumed that any QTL present may be anywhere within the maximum. The late owering line SL102 had a maximum GD in the region between 0.0 and 35.7 cm on chromosome O1, but the early owering line SL108 had a minimum between 30.3 and 35.7 cm, so it is unlikely that a late owering QTL was present in this region (Fig. 1, O1). It is therefore most likely that there was a late owering QTL between 0.0 and 30.3 cm and an early owering QTL between 30.3 and 38.1 cm on chromosome O1. Lines SL128 and SL133 both showed late owering behaviour. Assuming the presence of just one QTL on

5 590 A. M. RAE ET AL. Table 3 The mean owering date for the Brassica oleracea recurrent parent, A12, and the lines that showed signi cantly di erent owering behaviour from A12 over the ve years. Bold type denotes those lines that were signi cant and `Ð' indicates that the line was not raised in that year Line T T A SL Ð Ð SL SL112 Ð Ð Ð Ð Ð SL113 Ð Ð Ð Ð SL Ð Ð Ð Ð SL Ð Ð Ð Ð SL Ð Ð Ð Ð SL Ð Ð Ð Ð Ð SL128 Ð Ð Ð Ð Ð SL129 Ð Ð Ð Ð SL Ð Ð Ð Ð SL Ð Ð Ð Ð SL137 Ð Ð Ð Ð Ð SL Ð Ð Ð Ð SL140 Ð Ð Ð Ð SL Ð Ð Ð Ð SL Ð Ð Ð Ð SL157 Ð Ð Ð Ð Ð SL Ð Ð Ð Ð SL SL Ð SL Ð Ð Ð SL178 Ð Ð Ð Ð Ð chromosome O3, it is likely to be located in the region in which the s in these lines overlap, i.e. between 30.5 and 43.3 cm, although the presence of a second QTL in this region is discussed later (Fig. 1, O3). Line SL157 suggested the presence of a late owering QTL between 0.0 and 34.2 cm on chromosome O5 in GD (Fig. 1. O5). The ve late owering lines with single s on chromosome O9 suggest three separate regions to which QTL may be mapped. Lines SL142, SL175 and SL178 have s which overlap between 70.8 and cm; line SL172 has a between 0.0 and 43.0 cm; and line SL177 has a between 43.0 and 64.4 cm (Fig. 1, O9). If these seven QTL are assumed to be present in the regions described, inferences can now be made about those lines which contain more than one QTL. Lines with more than one Of the 13 signi cant lines with more than one GD, the behaviour of 11 can be explained by QTL found in lines with single s. Inferences for QTL in the two lines SL122 and SL127 cannot be made using line data alone, but are discussed below. Comparison to previous work Previous work carried out on this A12 GD cross by Bohuon et al. (1998) analysed data from eld trials using a mapping method based on marker regression (Kearsey & Hyne, 1994). This method showed evidence for six QTL over the two trials analysed in Single QTL were found on chromosomes O2 and O3, whereas chromosomes O5 and O9 both showed evidence for two QTL. Figure 1 provides a comparison of the doubled haploid and line data. The QTL are `named' according to their chromosome and position in the two doubled haploid trials simply to facilitate comparisons (see Bohuon et al., 1998). Chromosome O1 The doubled haploid data showed no evidence for QTL on chromosome O1, whereas the line data showed the presence of two QTL; one late owering between 0.0 and 30.3 cm and one early owering between 30.3 and 38.1 cm. As the regions to which these two putative QTLs map are adjacent to each other, they are likely to cancel each other in the doubled haploid lines. Chromosome O2 The QTL FTO2.1, which had a late owering allele from GD, was located at 78 9 cm on chromosome O2. Eight of the lines have introgressed GD DNA at this position. Of these, four owered signi cantly late, one owered early and three did not show signi cant variation from A12. The ve that showed signi cant owering variation all have s on other chromosomes which may be a ecting owering time. SL112 and SL113 are thought to contain a region coding for late owering on chromosome O3; SL121 contains late owering regions on chromosomes O1 and O9; and the early owering of SL122 may be caused by a QTL mapped in the region of FTO9.1 in the doubled haploid lines. It is possible that SL127 is re ecting the late owering e ect of FTO2.1. It may also be possible that lines SL112 and SL113 show this e ect as well as the late owering QTL on chromosome O3, because the mean owering dates for lines SL112 and SL113 were 17 and 16 days later than A12, respectively, in 1994, whereas the other signi cantly late lines in this year owered just 12±13 days later than A12.

6 QTL FOR FLOWERING TIME 591 Fig. 1 (see caption on p. 593)

7 592 A. M. RAE ET AL. Fig. 1 (see caption on p. 593)

8 QTL FOR FLOWERING TIME 593 Fig. 1 The positions of introgressed regions in the Brassica oleracea lines on chromosomes O1, O2, O3, O5 and O9. Minimum GD regions are shown as a box, whereas the maximum region, including where recombination occurred, is shown as a single line. The number of s contained by each line is noted. The signi cantly late owering lines are shaded in black, whereas those that were signi cantly early owering are shaded in grey. The regions to which the lines best map QTL are shown by vertical, dotted lines. For comparison, the regions to which the doubled haploid data mapped QTL are also shown. Chromosome O3 The late owering lines SL128 and SL133 have single s which overlap, suggesting that FTO3.1 is in the overlapping region, but the maximum in the late owering line SL115 only just overlaps the maximum in SL133. There is no evidence that the GD s on chromosome O1 or O6 in SL115 a ect the owering behaviour of this line, therefore there may be two QTL a ecting owering time on this chromosome. This view is supported by the size of the e ects of the QTL on this chromosome. The lines SL115 and SL133 owered 2 and 1.5 days later than the A12 parent, respectively, in 1998, whereas line SL134, which overlaps both regions, owered 6 days later than the A12 parent, implying the presence of two QTL in this line. The presence of a second QTL on chromosome O3 may explain the large con dence interval of FTO3.1 in the doubled haploid data. Chromosome O5 FTO5.1 maps to cm, whereas FTO5.2 maps to cm. The line data suggest that there is a QTL between 0.0 and 34.2 cm. This region overlaps with the con dence intervals for both FTO5.1 and FTO5.2, so the lines neither con rm nor reject the two-qtl hypothesis. Chromosome O9 The doubled haploid data indicated the presence of two QTL on chromosome O9; an early QTL (FTO9.1) at 46 cm in trial 1 and 36 cm in trial 2, and a late QTL (FTO9.2) at 74 cm in trial 1 and 94 cm in trial 2. The marker regression approach used on these doubled haploid data does not give con dence intervals when more than one QTL is mapped to a chromosome. The -line data map a late owering QTL between 70.8 and cm, which coincides with the doubled haploid QTL FTO9.2. The doubled haploid data did not reveal the two late QTL found in the lines in the regions of 0.0±43.0 cm and 43.0±64.4 cm, but did detect an early e ect at 46 cm in trial 1 and 36 cm in trial 2. This may be the cause of early owering in SL122, which has a GD between 23.1 and 47.0 cm on chromosome O9. The late owering line SL129 completely

9 594 A. M. RAE ET AL. overlaps the region to which this early owering QTL is mapped, but it is possible that this line contains the two late owering QTLs at this end of chromosome O9, which mask the e ect of early owering. There is some evidence that the doubled haploid lines show segregation distortion in favour of GD in the region of the early QTL, which may explain why the early QTL alone is detected in the doubled haploid data. Precision of mapping The doubled haploid data mapped QTL with 95% con dence intervals that varied between 18 and 52 cm. Usually a con dence interval of 30 cm is thought to be reasonable for analysis of segregating populations. In the lines the interval to which a QTL can be mapped relies on the size of the and also the distance between markers on the linkage map. Substitution lines that showed signi cant e ects for owering time contained introgressed GD DNA that ranged in length from 8 cm to 43 cm, showing that this method of mapping QTL does enable e ects to be mapped more precisely. The line approach identi ed at least 11 QTL compared to the six detected by the doubled haploids. Substitution lines identi ed more QTL, because they detected e ects that had been masked in the doubled haploid population through dispersion and close linkage. It is possible that there are still more QTL for owering time segregating in this cross. Some of the lines contained introgressed regions as long as 84 cm, so it is quite feasible that such large regions may contain more than one QTL, which mask each other's e ect. The lines grown over the ve years represent 62% of the entire GD genome, although the actual proportion may be larger because this calculation does not include the region within which recombination has occurred. Only 9.0% of the GD genome is de nitely not present in the lines. This implies that there may well be QTL in the regions that are not represented in the lines. Comparison to other genomes The use of the linkage map produced by Bohuon et al. (1996) in this study has enabled a comparison of results to be made. This map has also been used to compare the C genomes of B. oleracea and B. napus, and colinearity of the large segments of the A and C genomes in B. napus (Parkin et al., 1995). Field experiments involving Brassica napus populations have indicated that regions on linkage groups in the A genome which show homology to the regions in O2, O3 and O9 also carry QTL for owering time (Keith, 1996; Salinas-Garcia, 1996; Osborn et al., 1997). These regions have also been shown to be homologous with regions on LG2 and LG8 of B. nigra, to which a owering time QTL has been mapped (Lagercrantz et al., 1996) and to the region around the CONSTANS gene and several other owering candidate genes on chromosome 5 in A. thaliana. These regions of homology have been found to map to the regions of B. oleracea chromosomes O2, O3 and O9, and to show similar e ects of late owering in GD (Bohuon et al., 1998). These appear to be the same as the substituted regions in the present study that also showed evidence of QTL for late owering in GD. The line data showed similar results to the doubled haploid data, but were not as easily interpreted. To maximize the bene ts of the use of lines, it would be advantageous to backcross lines further to reduce the size and number of introgressed GD regions in each line, and to map more markers to the linkage map to enable reduction of the regions in which recombination is presumed to have occurred. Other measurements Measurements for height and number of nodes at owering were also taken in the eight line trials, as these have been found to have a positive correlation with owering time (Bohuon, 1996). Late owering tends to be a result of extended vegetative growth rather than stunted or delayed growth (Kowalski et al., 1994). Future work will involve the analysis of these traits and a comparison to owering time data. Acknowledgements The authors particularly thank Judith Craft and Sue Bradshaw for technical support. This work was supported by the UK Biotechnology and Biological Sciences Council (BBSRC) research grant number 6/G References BOHUON, E. J. R A Genetic Analysis of Brassica oleracea. Ph.D. Thesis, University of Birmingham. BOHUON, E. J. R., KEITH, D. J., PARKIN, I. A. P., SHARPE, A. G. AND LYDIATE, D. J. D. J Alignment of the conserved C genomes of Brassica oleracea and Brassica napus. Theor. Appl. Genet., 93, 833±839. BOHUON, E. J. R., RAMSAY, L. D., CRAFT, J. A., ARTHUR, A. E., MARSHALL, D. F., LYDIATE, D. J. ET AL. ET AL The association of owering time quantitative trait loci with duplicated regions and candidate loci in Brassica oleracea. Genetics, 150, 393±401. CAMARGO, L. E. A. AND OSBORN, T. C Mapping loci controlling owering time in Brassica oleracea. Theor. Appl. Genet., 92, 610±616.

10 QTL FOR FLOWERING TIME 595 DUNNETT, C. W Pairwise multiple comparisons in the homogeneous variance, unequal sample size case. J. Am. Stat. Ass., 75, 789±795. KEARSEY, M. J. AND HYNE, V QTL analysis: a simple `marker-regression' approach. Theor. Appl. Genet., 89, 698±702. KEARSEY, M. J. AND POONI, H. S The Genetical Analysis of Quantitative Traits. Chapman & Hall, London. KEITH, D. J Genetical Analysis of Quantitative Traits in Brassica napus. Ph.D. Thesis, University of East Anglia. KENNARD, W. C., SLOCUM, M. K., FIGDORE, S. S. AND OSBORN, T. C Genetic analysis of morphological variation in Brassica oleracea using molecular markers. Theor. Appl. Genet., 87, 721±732. M., HANHART, C. J. AND VAN DER VEEN, J. H A genetic and physiological analysis of late owering mutants in Arabidopsis thaliana. Mol. Gen. Genet., 229, 57±66. KOORNNEEF, M. KOWALSKI, S. P., LAN, T. H., FELDMANN, K. A. AND PATERSON, A. H. H QTL mapping of naturally-occurring variation in owering time in Arabidopsis thaliana. Mol. Gen. Genet., 245, 548±555. LYDIATE, D. J Comparative genome mapping in Brassica. Genetics, 144, 1903±1910. LAGERCRANTZ, U. AND LAGERCRANTZ, U., PUTTERILL, J., COUPLAND AND, G. AND LYDIATE, D Comparative mapping in Arabidopsis and Brassica, ne scale colinearity and congruence of genes controlling owering time. Plant J., 9, 13±20. L. A. AND SCARTH, R Vernalisation response in spring oilseed rape (Brassica napus L.) cultivars. Can. J. Plant Sci., 74, 275±277. MURPHY, L. A. OSBORN, T. C., KOLE, C., PARKIN, I. A. P., SHARPE, A. G., KUIPER, M., LYDIATE, D. J. ET AL. ET AL Comparison of owering time genes in Brassica rapa. B. napus and Arabidopsis thaliana. Genetics, 146, 1123±1129. PARKIN, I., SHARPE, A. G., KEITH, D. J. AND LYDIATE, D. J Identi cation of the A and C genomes of amphidiploid Brassica napus (oilseed rape). Genome, 38, 1122±1131. PUTTERILL, J., ROBSON, F., LEE, K. AND COUPLAND AND, G. G Chromosome walking with YAC clones in Arabidopsis: isolation of 1700 kb of contiguous DNA on chromosome 5, including a 300 kb region containing the owering time gene CO. Mol. Gen. Genet., 239, 145±157. PUTTERILL, J., ROBSON, F., LEE, K., SIMON, R. AND COUPLAND AND, G The CONSTANS gene of Arabidopsis promotes owering and encodes a protein showing similarities to zinc nger transcription factors. Cell, 80, 847±857. RAMSAY, L. D., JENNINGS, D. E., BOHUON, E. J. R., ARTHUR, A. E., LYDIATE, D. J., KEARSEY, M. J. ET AL. ET AL The construction of a library of recombinant backcross lines in Brassica oleracea for the precision mapping of quantitative trait loci. Genome, 39, 558±567. SALINAS-GARCIA, G. G Mapping Quantitative Trait Loci Controlling Traits in Brassica napus L. Ph.D. Thesis, University of Birmingham. SEBASTIAN, R. L., HOWELL, E. C., KING, G. J., MARSHALL, D. F. AND KEARSEY, M. J. M. J An integrated AFLP and RFLP Brassica oleracea linkage map from two morphologically distinct doubled haploid populations. Theor. Appl. Genet., in press. THURLING, N. AND DEPITTAYANAN, V. V EMS induction of early owering mutants in spring rape (Brassica napus). Pl. Breed., 108, 177±184. N Genome analysis in Brassica with special reference to the experimental formation of B. napus and peculiar mode of fertilisation. Jap. J. Bot., 7, 389±452. U, N. Appendix 1 Line Chromosome Minimum Maximum Line Chromosome Minimum Maximum SL100 Ð Ð Ð SL138 O3 35.9± ±54.2 SL101 O ± ±108.6 SL138 O5 0.0± ±34.2 SL102 O1 0.0± ±35.7 SL138 O ±20.2 SL103 O1 50.6± ±52.0 SL139 O3 43.3± ±72.1 SL104 O ±53.9 SL139 O ±100.1 SL105 O1 50.6± ±108.6 SL140 O ±130.5 SL106 O1 52.0± ±108.6 SL140 O8 73.2± ±79.3 SL107 O1 50.6± ±88.3 SL140 O9 55.5± ±106.8 SL108 O1 30.3± ±38.1 SL141 O ±116.2 SL109 O ±52.0 SL141 O ± ±128.7 SL110 O ±52.0 SL141 O8 68.4± ±79.3 SL111 O1 95.8± ±108.6 SL141 O9 53.2± ±106.8 SL111 O ±72.5 SL142 O9 69.1± ±106.8 SL112 O ±52.0 SL143 O4 0.0± ±14.9 SL112 O2 73.0± ±80.8 SL144 O4 68.4± ±100.1 SL112 O3 7.5± ±43.3 SL145 O ±100.1 SL113 O1 30.3± ±52.0 SL146 O ±100.1 SL113 O2 73.0± ±80.8 SL147 O4 68.6± ±100.1

11 596 A. M. RAE ET AL. Appendix 1 (Continued ) Line Chromosome Minimum Maximum Line Chromosome Minimum Maximum SL113 O3 7.5± ±67.2 SL148 O ±100.1 SL114 O1 41.9± ±88.3 SL149 O5 90.7± ±92.9 SL114 O3 60.2± ±78.9 SL150 O5 67.5± ±92.9 SL114 O ± ±140.1 SL151 O ±100.1 SL114 O6 0.0± ±14.8 SL151 O5 67.5± ±92.9 SL115 O ±52.0 SL152 O ±100.1 SL115 O3 22.5± ±35.9 SL152 O5 73.5± ±92.9 SL115 O ±20.2 SL153 O ±67.5 SL116 O1 38.1± ±52.0 SL154 O5 34.2± ±44.7 SL116 O3 22.5± ±35.9 SL155 O5 0.0± ±34.2 SL116 O ±20.2 SL155 O ±20.2 SL117 O ± ±108.6 SL156 O ±42.7 SL117 O4 0.0± ±68.6 SL157 O5 7.0± ±42.7 SL118 O1 95.8± ±108.6 SL158 O3 35.9± ±40.3 SL118 O ±72.5 SL158 O5 0.0± ±34.2 SL118 O7 5.9± ±36.5 SL158 O ±20.2 SL119 O1 95.8± ±108.6 SL159 O5 90.7± ±92.9 SL119 O ±116.2 SL159 O8 1.9± ±79.3 SL119 O7 5.9± ±15.7 SL160 O ±100.1 SL120 O ±116.2 SL160 O5 73.5± ±92.9 SL121 O ±30.3 SL160 O8 1.9± ±17.7 SL121 O ± ±116.2 SL161 O ±100.1 SL121 O9 13.8± ±43.0 SL161 O5 73.5± ±92.9 SL122 O2 88.4± ±116.2 SL161 O ±4.3 SL122 O6 2.1± ±18.0 SL162 O6 2.1± ±41.1 SL122 O ±47.0 SL162 O8 1.9± ±17.7 SL123 O2 73.0± ±80.8 SL163 O ±72.0 SL124 O ±41.7 SL164 O7 49.5± ±72.0 SL124 O ±72.0 SL165 O7 47.6± ±72.0 SL124 O ±106.8 SL165 O8 61.8± ±72.1 SL125 O ±41.7 SL166 O ±38.2 SL125 O ±72.0 SL167 O ±20.2 SL126 O ±41.7 SL167 O ±72.0 SL126 O7 49.5± ±72.0 SL168 O7 49.5± ±72.0 SL127 O2 70.2± ±80.8 SL169 O ±116.2 SL127 O4 39.5± ±100.1 SL169 O ±72.5 SL128 O3 22.5± ±43.3 SL169 O7 5.9± ±36.5 SL129 O ±116.2 SL170 O7 49.5± ±72.0 SL129 O ±58.1 SL171 O1 30.3± ±38.1 SL130 O ±72.1 SL171 O ±17.7 SL131 O3 60.2± ±76.1 SL172 O9 0.0± ±43.0 SL132 O3 43.3± ±72.1 SL173 O ± ±140.1 SL133 O3 35.9± ±76.1 SL173 O8 73.2± ±79.5 SL134 O ±116.2 SL173 O ±106.8 SL134 O3 0.0± ±76.1 SL174 O ±58.1 SL135 O3 22.5± ±72.1 SL175 O9 80.9± ±106.8 SL136 O3 43.3± ±67.2 SL176 O ±58.1 SL136 O ±100.1 SL177 O9 47.0± ±64.4 SL137 O ±54.2 SL178 O ±106.8 SL137 O5 7.0± ±42.7 SL179 O6 0.0± ±14.8